howto:xas_tdp
Differences
This shows you the differences between two versions of the page.
Both sides previous revisionPrevious revisionNext revision | Previous revision | ||
howto:xas_tdp [2021/08/02 15:13] – [Simple examples] abussy | howto:xas_tdp [2024/02/24 10:01] (current) – oschuett | ||
---|---|---|---|
Line 1: | Line 1: | ||
- | ====== How to run XAS LR-TDDFT calculations ====== | + | This page has been moved to: https://manual.cp2k.org/trunk/methods/properties/x-ray/tddft.html |
- | + | ||
- | This a a short tutorial on how to run near-edge X-ray absorption spectroscopy calculations using linear-response TDDFT. The method is implemented in CP2K under the XAS_TDP name. It relies on core-level specific approximations that enables efficient calculations of large and periodic systems. Both K- and L-edge are available. The details of the method can be found in [[ https:// | + | |
- | + | ||
- | ===== Brief theory recap ===== | + | |
- | + | ||
- | The method is based on 3 main core-specific approximations that boost the calculation efficiency. The first one is the core-valence separation. Due to large differences in energy and localization, | + | |
- | + | ||
- | + | ||
- | The second approximation is the sudden approximation, | + | |
- | + | ||
- | + | ||
- | Finally, a lot of 4-center 2-electron integrals have to be computed. Thanks to the core-valence separation and the sudden approximation, | + | |
- | + | ||
- | \begin{equation} | + | |
- | (pI|Jq) \approx \sum_{\mu, \nu} \ (pI|\mu) \ (\mu|\nu)^{-1} \ (\nu|Jq) | + | |
- | | + | |
- | + | ||
- | where $p,q$ represent atomic orbitals (Gaussian type orbitals/ | + | |
- | + | ||
- | For the exchange-correlation kernel, the RI scheme reads: | + | |
- | + | ||
- | \begin{equation} | + | |
- | (pI|f_{xc}|Jq) \approx \sum_{\kappa, | + | |
- | \end{equation} | + | |
- | + | ||
- | where all integrals but $(\lambda|f_{xc}|\mu)$ are the same as for the Coulomb kernel above. Since the RI basis elements $\lambda, \mu$ are centered on the excited atoms, we only need the density in its vicinity. For this, we use a simple projection: | + | |
- | + | ||
- | \begin{equation} \label{proj} | + | |
- | \begin{aligned} | + | |
- | n(\mathbf{r}) & | + | |
- | %\pause | + | |
- | & | + | |
- | &= \sum_\nu d_\nu \ \chi_\nu(\mathbf{r}), | + | |
- | \end{aligned} | + | |
- | \end{equation} | + | |
- | + | ||
- | which turns the density into a linear combination of RI basis elements. This allows for easy and simple numerical integration of $(\lambda|f_{xc}|\mu)$. Note that the quality of the projection may suffer if there are (heavy) atoms close by since their core states may not be well described (GTOs are only sharp at their center). This can be addressed by either using pseudopotentials for the neighbors or adding their RI basis function for the projection. | + | |
- | + | ||
- | For these approximations to work, the core states to be excited need to be identified among the Kohn-Sham oritals. They need to have a strong $1s$, $2s$ or $2p$ nature and be well localized. Not fullfilling these conditions will lead to wrong results. | + | |
- | + | ||
- | ===== The XAS_TDP input section ===== | + | |
- | + | ||
- | The parameters defining XAS LR-TDDFT calculations are found in the '' | + | |
- | + | ||
- | The most important keywords and subsections of '' | + | |
- | * '' | + | |
- | * '' | + | |
- | * '' | + | |
- | + | ||
- | The defaults value of all other keywords are in principle good enough. | + | |
- | + | ||
- | Note that the first requirement for XAS LR-TDDFT is that the ground state calculation on which it is based is of good quality. | + | |
- | ===== Simple examples ===== | + | |
- | + | ||
- | Illustrative examples usually tell more than long texts. Below, some typical input examples are displayed with explanations. They should cover most common use cases. | + | |
- | ==== CO$_2$ molecule (K-edge)==== | + | |
- | + | ||
- | This is a simple C and O K-edge calculation of the CO$_2$ molecule in the gas phase. The annotated input file is displayed below: | + | |
- | + | ||
- | <code - CO2.inp> | + | |
- | + | ||
- | & | + | |
- | PROJECT CO2 | + | |
- | RUN_TYPE ENERGY | + | |
- | &END GLOBAL | + | |
- | + | ||
- | & | + | |
- | &DFT | + | |
- | BASIS_SET_FILE_NAME EMSL_BASIS_SETS | + | |
- | POTENTIAL_FILE_NAME POTENTIAL | + | |
- | AUTO_BASIS RI_XAS MEDIUM | + | |
- | + | ||
- | & | + | |
- | CUTOFF 500 | + | |
- | REL_CUTOFF 40 | + | |
- | NGRIDS 5 | + | |
- | &END MGRID | + | |
- | + | ||
- | &QS | + | |
- | METHOD GAPW ! It is necesary to use the GAPW method for | + | |
- | &END QS ! accurate description of core states | + | |
- | + | ||
- | & | + | |
- | PERIODIC NONE | + | |
- | PSOLVER MT | + | |
- | &END | + | |
- | + | ||
- | &SCF | + | |
- | EPS_SCF 1.0E-8 | + | |
- | MAX_SCF 30 | + | |
- | &END SCF | + | |
- | + | ||
- | &XC | + | |
- | & | + | |
- | & | + | |
- | FUNCTIONAL HYB_GGA_XC_BHandHLYP | + | |
- | & | + | |
- | &END XC_FUNCTIONAL | + | |
- | &HF | + | |
- | | + | |
- | &END HF | + | |
- | &END XC | + | |
- | + | ||
- | & | + | |
- | & | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | &END DONOR_STATES | + | |
- | + | ||
- | GRID C 250 500 ! Integration grid dimensions for C and O excited atoms | + | |
- | GRID O 250 500 ! there are 250 angular points (Lebedev grid) and 500 | + | |
- | ! radial points | + | |
- | + | ||
- | & | + | |
- | | + | |
- | ! sphere radius around the excited atom for the density projection | + | |
- | & | + | |
- | &LIBXC | + | |
- | | + | |
- | &END LIBXC | + | |
- | & | + | |
- | & | + | |
- | FRACTION 0.5 ! Definition of the functional for the TDDFT kernel | + | |
- | & | + | |
- | &END KERNEL | + | |
- | &END XAS_TDP | + | |
- | + | ||
- | &END DFT | + | |
- | & | + | |
- | &CELL | + | |
- | ABC 10.0 10.0 10.0 | + | |
- | PERIODIC NONE | + | |
- | &END CELL | + | |
- | & | + | |
- | C 5.00 5.00 5.00 | + | |
- | O 5.00 5.00 6.16 | + | |
- | O 5.00 5.00 3.84 | + | |
- | &END COORD | + | |
- | &KIND C | + | |
- | BASIS_SET 6-311G** | + | |
- | POTENTIAL ALL ! for the correct description of core states | + | |
- | &END KIND | + | |
- | &KIND O | + | |
- | BASIS_SET 6-311G** | + | |
- | POTENTIAL ALL | + | |
- | &END KIND | + | |
- | &END SUBSYS | + | |
- | &END FORCE_EVAL | + | |
- | + | ||
- | + | ||
- | </ | + | |
- | + | ||
- | There are a few points of particular interest in the '' | + | |
- | + | ||
- | In the '' | + | |
- | + | ||
- | In the resulting output file, there is a '' | + | |
- | + | ||
- | <code cp2k> | + | |
- | + | ||
- | # Start of calculations for donor state of type 1s for atom 1 of kind C | + | |
- | + | ||
- | The following localized MO(s) have been associated with the donor state(s) | + | |
- | based on the overlap with the components of a minimal STO basis: | + | |
- | | + | |
- | 1 3 1.00113 | + | |
- | + | ||
- | The next best overlap for spin 1 is 0.00000 for MO with index 1 | + | |
- | + | ||
- | Mulliken population analysis retricted | + | |
- | Spin MO index | + | |
- | | + | |
- | </ | + | |
- | + | ||
- | + | ||
- | Both the overlap and the Mulliken charge should be as close to 1.0 as possible. This ensure that the molecular orbital selected is of the correct type (here projection on a C 1s Slater type orbital) and properly localized (there is a full electron associated to this MO, on this atom). If those numbers are lower, something went wrong with the core level identification. This is usually solved by increasing '' | + | |
- | + | ||
- | Spectral information are given in a separate file named '' | + | |
- | ==== Tetrahedral NaAlO$_2$ (K-edge, periodic) ==== | + | |
- | + | ||
- | This example is about crystalline sodium aluminate and illustrates how large periodic structures can be efficiently simulated. | + | |
- | + | ||
- | <code - sodal.inp> | + | |
- | & | + | |
- | | + | |
- | | + | |
- | | + | |
- | &END GLOBAL | + | |
- | + | ||
- | & | + | |
- | | + | |
- | & | + | |
- | BASIS_SET_FILE_NAME | + | |
- | ! the pcseg-n and admm-n basis set families can be downloaded at https://www.basissetexchange.org | + | |
- | BASIS_SET_FILE_NAME | + | |
- | BASIS_SET_FILE_NAME | + | |
- | POTENTIAL_FILE_NAME | + | |
- | AUTO_BASIS RI_XAS MEDIUM | + | |
- | + | ||
- | &QS | + | |
- | | + | |
- | &END QS | + | |
- | + | ||
- | & | + | |
- | | + | |
- | &END AUXILIARY_DENSITY_MATRIX_METHOD | + | |
- | + | ||
- | &SCF | + | |
- | | + | |
- | | + | |
- | + | ||
- | & | + | |
- | | + | |
- | | + | |
- | & | + | |
- | & | + | |
- | MAX_SCF | + | |
- | EPS_SCF | + | |
- | & | + | |
- | &END SCF | + | |
- | + | ||
- | & | + | |
- | | + | |
- | | + | |
- | | + | |
- | &END | + | |
- | + | ||
- | &XC | + | |
- | & | + | |
- | & | + | |
- | | + | |
- | &END | + | |
- | & | + | |
- | + | ||
- | & | + | |
- | FRACTION 0.45 | + | |
- | & | + | |
- | | + | |
- | | + | |
- | &END INTERACTION_POTENTIAL | + | |
- | & | + | |
- | | + | |
- | &END SCREENING | + | |
- | & | + | |
- | &END XC | + | |
- | + | ||
- | & | + | |
- | & | + | |
- | DEFINE_EXCITED BY_KIND | + | |
- | KIND_LIST Alx ! There is only one Alx atom in the coordinates since all Al | + | |
- | STATE_TYPES 1s ! atoms are equivalent under symmetry. The Alx atom is the only | + | |
- | N_SEARCH 1 ! one decribed at all-electron level, which is why we use | + | |
- | & | + | |
- | + | ||
- | | + | |
- | GRID Alx 150 300 | + | |
- | | + | |
- | ! up to 20.0 eV above the first energy | + | |
- | & | + | |
- | MINIMIZER DIIS ! The iterative OT solver is typically more efficient than | + | |
- | EPS_ITER 1.0E-4 | + | |
- | & | + | |
- | + | ||
- | & | + | |
- | & | + | |
- | & | + | |
- | SCALE_X 0.55 | + | |
- | & | + | |
- | &END XC_FUNCTIONAL | + | |
- | + | ||
- | & | + | |
- | | + | |
- | | + | |
- | SCALE 0.45 | + | |
- | &END EXACT_EXCHANGE | + | |
- | & | + | |
- | &END XAS_TDP | + | |
- | & | + | |
- | + | ||
- | & | + | |
- | &CELL | + | |
- | ABC 10.467947 | + | |
- | &END CELL | + | |
- | & | + | |
- | | + | |
- | | + | |
- | &END TOPOLOGY | + | |
- | &KIND O | + | |
- | BASIS_SET DZVP-MOLOPT-SR-GTH | + | |
- | BASIS_SET AUX_FIT FIT3 | + | |
- | POTENTIAL GTH-PBE | + | |
- | &END KIND | + | |
- | &KIND Na | + | |
- | ELEMENT Na | + | |
- | BASIS_SET DZVP-MOLOPT-SR-GTH | + | |
- | BASIS_SET AUX_FIT FIT3 | + | |
- | POTENTIAL GTH-PBE | + | |
- | &END | + | |
- | &KIND Al | + | |
- | BASIS_SET DZVP-MOLOPT-SR-GTH | + | |
- | BASIS_SET AUX_FIT FIT3 | + | |
- | POTENTIAL GTH-PBE | + | |
- | &END | + | |
- | &KIND Alx ! All atoms but the single Alx are described using pseudopotentials | + | |
- | ELEMENT Al ! This greatly reduces the number of basis function and the cost of | + | |
- | BASIS_SET pcseg-2 | + | |
- | BASIS_SET AUX_FIT admm-2 | + | |
- | POTENTIAL ALL | + | |
- | &END | + | |
- | & | + | |
- | &END FORCE_EVAL | + | |
- | + | ||
- | + | ||
- | </code> | + | |
- | + | ||
- | There are many performance oriented keywords and subsection in the above input. Most importantly, | + | |
- | + | ||
- | This input file would generate a spectrum such as the one visible on figure 4 of the [[ https://pubs.rsc.org/en/ | + | |
- | + | ||
- | ==== TiCl$_4$ molecule (L-edge + SOC) ==== | + | |
- | + | ||
- | This example covers L-edge spectroscopy with the addition of spin-orbit coupling. | + | |
- | + | ||
- | <code - TiCl4.inp> | + | |
- | + | ||
- | & | + | |
- | PROJECT TiCl4 | + | |
- | PRINT_LEVEL LOW | + | |
- | RUN_TYPE ENERGY | + | |
- | &END GLOBAL | + | |
- | & | + | |
- | &DFT | + | |
- | BASIS_SET_FILE_NAME | + | |
- | POTENTIAL_FILE_NAME | + | |
- | AUTO_BASIS RI_XAS | + | |
- | + | ||
- | & | + | |
- | PERIODIC NONE | + | |
- | PSOLVER MT | + | |
- | &END POISSON | + | |
- | &QS | + | |
- | METHOD GAPW | + | |
- | &END QS | + | |
- | + | ||
- | & | + | |
- | CUTOFF 800 | + | |
- | REL_CUTOFF 50 | + | |
- | NGRIDS 5 | + | |
- | &END | + | |
- | + | ||
- | &SCF | + | |
- | EPS_SCF 1.0E-8 | + | |
- | MAX_SCF 200 | + | |
- | & | + | |
- | | + | |
- | ALPHA 0.2 | + | |
- | BETA 1.5 | + | |
- | | + | |
- | &END MIXING | + | |
- | &END SCF | + | |
- | + | ||
- | &XC | + | |
- | & | + | |
- | & | + | |
- | FUNCTIONAL HYB_GGA_XC_B3LYP | + | |
- | & | + | |
- | &END XC_FUNCTIONAL | + | |
- | &HF | + | |
- | | + | |
- | &END HF | + | |
- | &END XC | + | |
- | + | ||
- | & | + | |
- | & | + | |
- | | + | |
- | | + | |
- | | + | |
- | &END DONOR_STATES | + | |
- | + | ||
- | TAMM_DANCOFF FALSE ! TDA is on by default, get full TDDFT like this | + | |
- | DIPOLE_FORM LENGTH | + | |
- | + | ||
- | GRID Ti 500 1000 ! This is a fairly dense grid | + | |
- | + | ||
- | EXCITATIONS RCS_SINGLET | + | |
- | EXCITATIONS RCS_TRIPLET | + | |
- | SOC ! with the SOC hamiltonian | + | |
- | + | ||
- | & | + | |
- | | + | |
- | & | + | |
- | & | + | |
- | FUNCTIONAL HYB_GGA_XC_B3LYP | + | |
- | & | + | |
- | &END XC_FUNCTIONAL | + | |
- | & | + | |
- | FRACTION 0.2 | + | |
- | & | + | |
- | &END KERNEL | + | |
- | &END XAS_TDP | + | |
- | + | ||
- | &END DFT | + | |
- | & | + | |
- | &KIND Cl | + | |
- | BASIS_SET def2-TZVPD | + | |
- | POTENTIAL ALL | + | |
- | RADIAL_GRID 80 ! The GAPW grids are also used to evaluate the SOC operator | + | |
- | LEBEDEV_GRID 120 ! it is good practice to use sligthly larger ones than the default | + | |
- | &END KIND | + | |
- | &KIND Ti | + | |
- | BASIS_SET def2-TZVPD | + | |
- | POTENTIAL ALL | + | |
- | RADIAL_GRID 80 | + | |
- | LEBEDEV_GRID 120 | + | |
- | &END KIND | + | |
- | &CELL | + | |
- | ABC 10.0 10.0 10.0 | + | |
- | PERIODIC NONE | + | |
- | &END CELL | + | |
- | & | + | |
- | COORD_FILE_FORMAT XYZ | + | |
- | COORD_FILE_NAME TiCl4.xyz | + | |
- | & | + | |
- | &END CENTER_COORDINATES | + | |
- | &END TOPOLOGY | + | |
- | &END SUBSYS | + | |
- | &END FORCE_EVAL | + | |
- | + | ||
- | </ | + | |
- | + | ||
- | The structure of the input file is not very different from the CO$_2$ example. Notable differences are the '' | + | |
- | + | ||
- | Note that this calculation is meant to be a benchmark calculation, | + | |
- | + | ||
- | + | ||
- | In the output file, the donor state identification yields overlaps that are greater than one. This is due to the degenerate nature of 2p states. The candidate Kohn-Sham orbital is projected on 3 STOs for 2px, 2py and 2pz. To avoid cancelling contributions, | + | |
- | + | ||
- | < | + | |
- | # Start of calculations for donor state of type 2p for atom 1 of kind Ti | + | |
- | + | ||
- | The following canonical MO(s) have been associated with the donor state(s) | + | |
- | based on the overlap with the components of a minimal STO basis: | + | |
- | | + | |
- | 1 7 1.36751 | + | |
- | 1 8 1.36751 | + | |
- | 1 9 0.99786 | + | |
- | + | ||
- | The next best overlap for spin 1 is 0.06653 for MO with index 27 | + | |
- | + | ||
- | Mulliken population analysis retricted to the associated MO(s) yields: | + | |
- | Spin MO index | + | |
- | | + | |
- | | + | |
- | | + | |
- | </ | + | |
- | + | ||
- | ===== FAQ ===== | + | |
- | + | ||
- | ==== Which functional and basis sets to use ? ==== | + | |
- | + | ||
- | Hybrid functionals with high fraction of Hartree-Fock exchange are know to perform well for core spectroscopy. PBEh($\alpha=0.45$) and BHandHLYP have had success with this particular implementation. In periodic boundary conditions, the truncated Coulomb operator should be used (with truncation radius < half cell parameter). | + | |
- | + | ||
- | For appropriate description of core states, all-electron basis sets should be used for the excited atom(s). MOLOPT basis sets and pseudopotentials can be used on all other atoms. There exist core specific basis such as pcX-n and cc-pCVXZ, but their usage is not necessary (based on basis set convergence studies on small molecules). Note that the pcseg-n basis sets are nice to use as they come with their own ADMM basis. | + | |
- | ==== How do I make my calculation more accurate ? ==== | + | |
- | + | ||
- | + | ||
- | The first necessity is to have a good ground state calculation. Thus, any change of paramter improving the '' | + | |
- | + | ||
- | Within '' | + | |
- | * '' | + | |
- | * Increasing the '' | + | |
- | * Increasing the '' | + | |
- | + | ||
- | By default, the RI basis used for the integral and the projection is autamatically generated. The quality of the RI basis can be changed via the '' | + | |
- | ==== How do I make my calculation faster ? ==== | + | |
- | + | ||
- | All points mentioned above in the accuracy section can also be tweaked for performence. In general, lowering accuracy will lead to faster calculations. | + | |
- | + | ||
- | For large systems, it is recommanded to use the '' | + | |
- | + | ||
- | The Tamm-Dancoff approximation is well established and generally yields results as good as full TDDFT. It is moreover much cheaper than the latter. It is turned on by default, but you may want to make sure it is enabled. | + | |
- | + | ||
- | The use of ADMM is highly recommanded for large systems, where the ground state HFX evaluation is the main bottleneck. It is also recommanded to use pseudopotentials on all atoms that are not excited as all-electron basis set tend to be large. If there exist no proper ADMM basis for the all-electron basis used for the excited atom, you may use the full basis as '' | + | |
- | + | ||
- | The code is also both MPI and OMP parallelized. Using more core will, to a certain degree, speedup your calculations as well. | + | |
- | ==== How do I plot a spectrum from the *.spectrum output file ==== | + | |
- | + | ||
- | For each donor state in the system, the *.spectrum file contains a list of excitation energies and corresponding oscillator strengths. This yields a stick spectrum which needs to be artificially broadened to match experiments. This is typically done using Gaussian of Lorentzian functions. Note that in case of spin-orbit calculation at the L-edge, results for the singlet, triplet and SOC excitations are given. | + | |
- | + | ||
- | Remember that XAS LR-TDDFT produces an accurate spectrum, but it is usually wrongly positioned on the energy axis. Again, to match experiment, a rigid shift must be applied to the result. | + | |
- | + | ||
- | + | ||
- | + | ||
- | ==== My calculation yields a wrong/ | + | |
- | + | ||
- | Assuming that the ground state calculation is sound and well converged, there are two main causes for failure. | + | |
- | + | ||
- | The proper core state was not found/ | + | |
- | + | ||
- | The numerical integration of the XC kernel $(\lambda|f_{xc}|\mu)$ is not accurate enough. In this case, you may want to increase the density of the integration grid with the '' | + | |
- | + | ||
- | Finally, keep in mind that calculated spectra need to be rigidly shifted by some energy to match experiment. | + | |
- | + | ||
- | ====== First-principles correction scheme ====== | + | |
- | As mentioned above, XAS LR-TDDFT results need to be rigidly shifted to match experiments. This is due to self-interaction error and the lack of orbital relaxation upon the creation of the core hole. An // | + | |
- | ==== Brief theory recap ==== | + | |
- | + | ||
- | XAS LR-TDDFT yield excitation energies as correction to ground state Kohn-Sham orbital energy differences, | + | |
- | + | ||
- | $$ | + | |
- | \omega = \varepsilon_a - \varepsilon_I + \Delta_{xc}, | + | |
- | $$ | + | |
- | + | ||
- | where $\varepsilon_a$ is the orbital energy of a virtual MO and $\varepsilon_I$ the energy of the donor core MO. Under Koopman' | + | |
- | + | ||
- | The IP can be accurately calculated using the second-order electron propagator equation: | + | |
- | + | ||
- | $$ | + | |
- | \text{IP}_I = -\varepsilon_I - \frac{1}{2} \sum_{ajk}\frac{|\langle Ia||jk\rangle|^2}{-\text{IP}_I + \varepsilon_a -\varepsilon_j -\varepsilon_k} - \frac{1}{2}\sum_{abj}\frac{|\langle Ij||ab\rangle|^2}{-\text{IP}_I + \varepsilon_j - \varepsilon_a - \varepsilon_b} | + | |
- | $$ | + | |
- | + | ||
- | where $a, b$ refer to virtual Hartree-Fock spin-orbitals and $j,k$ to occupied HF spin-orbitals. The DFT generalization of this theory is known as [[https:// | + | |
- | ==== The GW2X input subsection ==== | + | |
- | + | ||
- | The parameters defining the GW2X correction to XAS LR-TDDFT are found in the '' | + | |
- | + | ||
- | There are not many parameters to set for the GW2X correction. Simply adding an empty ''& | + | |
- | ==== Simple examples ==== | + | |
- | + | ||
- | === OCS molecule (L-edge + SOC) === | + | |
- | + | ||
- | This example covers GW2X corrected L-edge spectroscopy with spin-orbit coupling. | + | |
- | + | ||
- | <code - OCS.inp> | + | |
- | + | ||
- | & | + | |
- | PROJECT OCS | + | |
- | PRINT_LEVEL MEDIUM | + | |
- | RUN_TYPE ENERGY | + | |
- | &END GLOBAL | + | |
- | & | + | |
- | METHOD Quickstep | + | |
- | &DFT | + | |
- | BASIS_SET_FILE_NAME BASIS_GW2X | + | |
- | POTENTIAL_FILE_NAME POTENTIAL | + | |
- | AUTO_BASIS RI_XAS MEDIUM | + | |
- | + | ||
- | & | + | |
- | CUTOFF 800 | + | |
- | REL_CUTOFF 50 | + | |
- | NGRIDS 5 | + | |
- | &END MGRID | + | |
- | &QS | + | |
- | METHOD GAPW | + | |
- | &END QS | + | |
- | + | ||
- | & | + | |
- | PERIODIC NONE | + | |
- | PSOLVER MT | + | |
- | &END | + | |
- | + | ||
- | &SCF | + | |
- | EPS_SCF 1.0E-8 | + | |
- | MAX_SCF 50 | + | |
- | &END SCF | + | |
- | + | ||
- | &XC | + | |
- | & | + | |
- | & | + | |
- | FUNCTIONAL GGA_C_PBE | + | |
- | & | + | |
- | & | + | |
- | FUNCTIONAL GGA_X_PBE | + | |
- | SCALE 0.55 | + | |
- | & | + | |
- | &END XC_FUNCTIONAL | + | |
- | + | ||
- | &HF | + | |
- | | + | |
- | &END HF | + | |
- | &END XC | + | |
- | + | ||
- | & | + | |
- | & | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | &END DONOR_STATES | + | |
- | + | ||
- | EXCITATIONS RCS_SINGLET | + | |
- | EXCITATIONS RCS_TRIPLET | + | |
- | SOC | + | |
- | + | ||
- | GRID S 300 500 | + | |
- | + | ||
- | N_EXCITED 150 | + | |
- | TAMM_DANCOFF | + | |
- | + | ||
- | & | + | |
- | &END GW2X ! standard XAS_TDP calculation (defaults parameters are used) | + | |
- | + | ||
- | & | + | |
- | | + | |
- | & | + | |
- | &LIBXC | + | |
- | | + | |
- | &END LIBXC | + | |
- | &LIBXC | + | |
- | | + | |
- | SCALE 0.55 | + | |
- | &END LIBXC | + | |
- | & | + | |
- | & | + | |
- | FRACTION 0.45 | + | |
- | & | + | |
- | &END KERNEL | + | |
- | + | ||
- | &END XAS_TDP | + | |
- | &END DFT | + | |
- | & | + | |
- | &CELL | + | |
- | ABC 10.0 10.0 10.0 | + | |
- | PERIODIC NONE | + | |
- | &END CELL | + | |
- | & | + | |
- | C | + | |
- | O | + | |
- | S | + | |
- | &END COORD | + | |
- | &KIND C | + | |
- | BASIS_SET aug-pcX-2 | + | |
- | POTENTIAL ALL | + | |
- | &END KIND | + | |
- | &KIND O | + | |
- | BASIS_SET aug-pcX-2 | + | |
- | POTENTIAL ALL | + | |
- | &END KIND | + | |
- | &KIND S | + | |
- | BASIS_SET aug-pcX-2 | + | |
- | POTENTIAL ALL | + | |
- | &END KIND | + | |
- | &END SUBSYS | + | |
- | &END FORCE_EVAL | + | |
- | + | ||
- | </ | + | |
- | + | ||
- | The only difference between the above input file and that of a standard XAS LR-TDDFT calculation is the addition of the ''& | + | |
- | + | ||
- | In the output file, the correction for each S $2p$ is displayed. Note that the correction amounts to a shift of 1.9 eV compared to standard XAS LR-TDDFT, leading to a first singlet excitation energy of 164.4 eV (at the L$_3$ edge). This fits [[https:// | + | |
- | + | ||
- | < | + | |
- | - GW2X correction for donor MO with spin 1 and MO index 5: | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | + | ||
- | Final GW2X shift for this donor MO (eV): | + | |
- | + | ||
- | + | ||
- | - GW2X correction for donor MO with spin 1 and MO index 6: | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | + | ||
- | Final GW2X shift for this donor MO (eV): | + | |
- | + | ||
- | + | ||
- | - GW2X correction for donor MO with spin 1 and MO index 7: | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | + | ||
- | Final GW2X shift for this donor MO (eV): | + | |
- | + | ||
- | + | ||
- | Calculations done: | + | |
- | + | ||
- | First singlet XAS excitation energy (eV): 165.014087 | + | |
- | First triplet XAS excitation energy (eV): 164.681850 | + | |
- | First SOC XAS excitation energy (eV): 164.396537 | + | |
- | + | ||
- | Ionization potentials for XPS (GW2X + SOC): 170.602279 | + | |
- | | + | |
- | | + | |
- | + | ||
- | </ | + | |
- | + | ||
- | === Solid NH$_3$ (K-edge, periodic) === | + | |
- | + | ||
- | This is a much larger example of a periodic system, namely solid ammonia. This example is much heavier to run (~45 minutes on 24 cores). | + | |
- | + | ||
- | <code NH3.inp> | + | |
- | + | ||
- | & | + | |
- | PROJECT NH3 | + | |
- | RUN_TYPE ENERGY | + | |
- | PRINT_LEVEL MEDIUM | + | |
- | &END GLOBAL | + | |
- | & | + | |
- | METHOD QS | + | |
- | &DFT | + | |
- | BASIS_SET_FILE_NAME BASIS_PCSEG | + | |
- | BASIS_SET_FILE_NAME BASIS_ADMM | + | |
- | BASIS_SET_FILE_NAME BASIS_MOLOPT | + | |
- | POTENTIAL_FILE_NAME POTENTIAL | + | |
- | AUTO_BASIS RI_XAS MEDIUM | + | |
- | + | ||
- | &QS | + | |
- | METHOD GAPW | + | |
- | &END QS | + | |
- | + | ||
- | & | + | |
- | CUTOFF 600 | + | |
- | REL_CUTOFF 50 | + | |
- | NGRIDS 5 | + | |
- | &END MGRID | + | |
- | + | ||
- | &SCF | + | |
- | SCF_GUESS RESTART | + | |
- | EPS_SCF 1.0E-8 | + | |
- | MAX_SCF 30 | + | |
- | + | ||
- | &OT | + | |
- | | + | |
- | | + | |
- | &END OT | + | |
- | + | ||
- | & | + | |
- | | + | |
- | | + | |
- | &END OUTER_SCF | + | |
- | + | ||
- | &END SCF | + | |
- | + | ||
- | & | + | |
- | ADMM_PURIFICATION_METHOD NONE | + | |
- | &END AUXILIARY_DENSITY_MATRIX_METHOD | + | |
- | + | ||
- | &XC | + | |
- | & | + | |
- | & | + | |
- | FUNCTIONAL GGA_X_PBE | + | |
- | SCALE 0.55 | + | |
- | & | + | |
- | & | + | |
- | FUNCTIONAL GGA_C_PBE | + | |
- | & | + | |
- | &END XC_FUNCTIONAL | + | |
- | &HF | + | |
- | | + | |
- | & | + | |
- | POTENTIAL_TYPE TRUNCATED | + | |
- | CUTOFF_RADIUS 5.0 | + | |
- | & | + | |
- | &END HF | + | |
- | &END XC | + | |
- | + | ||
- | & | + | |
- | & | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | | + | |
- | &END DONOR_STATES | + | |
- | + | ||
- | TAMM_DANCOFF | + | |
- | GRID Nx 300 500 | + | |
- | E_RANGE 30.0 | + | |
- | + | ||
- | &GW2X | + | |
- | &END | + | |
- | + | ||
- | & | + | |
- | & | + | |
- | & | + | |
- | | + | |
- | SCALE 0.55 | + | |
- | &END | + | |
- | & | + | |
- | | + | |
- | &END | + | |
- | & | + | |
- | & | + | |
- | OPERATOR TRUNCATED | + | |
- | CUTOFF_RADIUS 5.0 | + | |
- | FRACTION 0.45 | + | |
- | & | + | |
- | &END KERNEL | + | |
- | + | ||
- | &END XAS_TDP | + | |
- | &END DFT | + | |
- | & | + | |
- | &CELL | + | |
- | ABC | + | |
- | &END CELL | + | |
- | & | + | |
- | COORD_FILE_FORMAT XYZ | + | |
- | COORD_FILE_NAME NH3.xyz | + | |
- | &END TOPOLOGY | + | |
- | &KIND H | + | |
- | BASIS_SET DZVP-MOLOPT-SR-GTH | + | |
- | BASIS_SET AUX_FIT FIT3 | + | |
- | POTENTIAL GTH-PBE | + | |
- | &END KIND | + | |
- | &KIND N | + | |
- | BASIS_SET DZVP-MOLOPT-SR-GTH | + | |
- | BASIS_SET AUX_FIT FIT3 | + | |
- | POTENTIAL GTH-PBE | + | |
- | &END KIND | + | |
- | &KIND Nx | + | |
- | ELEMENT N | + | |
- | BASIS_SET aug-pcseg-2 | + | |
- | BASIS_SET AUX_FIT aug-admm-2 | + | |
- | POTENTIAL ALL | + | |
- | &END KIND | + | |
- | &END SUBSYS | + | |
- | &END FORCE_EVAL | + | |
- | + | ||
- | </ | + | |
- | + | ||
- | Again, the only difference with respect to a standard XAS-LRTDDFT input file is the ''& | + | |
- | + | ||
- | + | ||
- | ==== FAQ ==== | + | |
- | + | ||
- | === How can I make the GW2X correction run faster ?=== | + | |
- | + | ||
- | === Why don't I get the absolute core IP in periodic systems ? === | + | |
- | + | ||
- | For molecules in non-periodic boundary conditions, the potential is such that it is zero far away. In the periodic case, the zero is ill defined. As a consequence, | + |
howto/xas_tdp.1627917215.txt.gz · Last modified: 2021/08/02 15:13 by abussy